Conventionally, in digital still cameras, video cameras, and so forth, in order to prevent image degradation due to cancellation of digits in digital signal processing, the grayscale range of the signal at the input and processing stages (around 10 to 12 bits) is set to be larger than the grayscale range of the final output signal (normally 8 bits). In this case, it is necessary to perform grayscale conversion so as to match the grayscale range of the output system. Conventionally, grayscale conversion is performed with static grayscale characteristics for a reference scene. In addition, a method has also been proposed in which the grayscale characteristics are obtained for an acquired scene, and the grayscale is converted adaptively.
For example, Japanese Unexamined Patent Application, Publication No. 2003-69821 (Patent Document 1) discloses a technique for performing grayscale conversion in which importance is placed on the main subject by estimating the image acquisition conditions. Patent Document 1 also discloses a technique for suppressing the adverse effects of noise and so forth by imposing limits on the obtained grayscale characteristics.
On the other hand, a method of performing grayscale conversion by independently changing the grayscale conversion characteristics for each region in an image signal (space-variant) has also been proposed. For example, Publication of Japanese Patent No. 3465226 discloses a technique for dividing a video signal into regions on the basis of texture information and performing adaptive grayscale conversion for each region (in other words, a technique for performing space-variant grayscale conversion).
Patent Document 1:
    Japanese Unexamined Patent Application, Publication No. 2003-69821Patent Document 2:    Publication of Japanese Patent No. 3465226